ProgrammableMarketing/width.ai (opens in a new tab)
PDF

Views

  • Plan
  • Roadmap
  • Research

On this page

  • The brief
  • Start here
  • Idea portfolio
  • Effort vs impact
  • This week · 9
  • This quarter · 11
  • Big bets · 3
  • Wildcards · 1
Generated Nov 19, 2026
  1. Programmable
  2. Playbooks
  3. Width.ai
Width.ai logo

growth playbook · Nov 19, 2026

width.ai (opens in a new tab)

Width.ai growth playbook

Diagnosis

Width.ai does current, credible work: a 14-agent listing system for a holding company serving 3,000 brands, a voice agent used by 10,000+ customers, 89-97% accuracy catalogue and SKU studies from 2024-2026, and a weekly stream of n8n and agent posts. The website hides almost all of it. The /about page is still an unedited Webflow template for a fictional company called "Wonder", with Lorem Ipsum and a $55M funding timeline.

The homepage leads with a keyword-list H1 and 2021-2023 case studies, shows a stray testimonial for another product three times, and has header and blog links that go nowhere. The best proof sits on one services page, the free n8n workflows capture no leads, and the contact form asks nothing that would qualify a project.

This playbook fixes the trust pages first (about, homepage proof, H1, contact form, robots.txt, template leftovers), then turns the n8n audience into pipeline with a service page, a published POC price band, a cost guide and a workflow library, gives Pumice a self-serve sample, and builds recurring revenue through evals, retainers and partner co-selling.

Growth Score

6/10

Biggest Opportunity

Put the 2024-2026 agentic and catalogue work in front of the buyers already reading the n8n and agent content, and give them a clear, priced next step.

Biggest Risk

A buyer doing due diligence finds a template about page, borrowed testimonials and dead links, and picks one of the many AI agencies (LeewayHertz, Markovate, Addepto and freelancers) whose sites look finished.

01

Start here

The three moves to make first. Each one is written up in full below.

  1. 1

    Replace the /about page: it is still an unedited Webflow template for a company called "Wonder"

    Unpublish the template sections today and put up a three-paragraph about page with the founder, the focus and three numbers.

    High impact First result in 1 week
  2. 2

    Lead the homepage with the 2024-2026 agentic and catalogue work instead of 2021-2023 case studies

    Pick the six results and write a 12-word caption for each.

    High impact First result in 2 weeks
  3. 3

    Rewrite the homepage H1 from a keyword list into the offer: production AI agents, with a POC in 31 days

    Write three H1 options and pick one with the founder.

    Medium impact First result in 2 weeks

02

Your idea portfolio

24 ideas, grouped by when to run them. Open any idea for how to run it: the steps, who to approach, copy you can use, and how to measure it.

  • 8 ideas are marked Unexpected: plays you are unlikely to have tried.

Every idea placed by the effort it takes against the impact it could have. Quick wins sit top left; select a number to open that idea. The filters below narrow the list by status, timing, category and what Programmable can build.

Effort vs impactEach number is an idea. Select one to open it.
Impact →
Quick wins · 3Big bets · 8Fill-ins · 8Reconsider · 5
Effort →

Showing 24 of 24 ideas

This week 9

Quick wins with a clear first action.

  • 01

    width.ai/about, the page a buyer opens before trusting a consultancy with their data, is the untouched "Wonder - Webflow HTML Website Template". It says "We're a team of creators and innovators", tells a story about helping startups "with beautifully designed templates", lists "250,000 paying customer", "Wonder's IPO", $55M raised and a 12-person team led by a placeholder founder, all in Lorem Ipsum. Width appears only in the menu and footer. The page is in the sitemap, so search engines, AI assistants and company-data tools read it too: company data built from width.ai already lists the template's placeholder CFO and CMO as Width's leadership.

    High impactLow effort First result in 1 week Founder & Chief Architect

    First action

    Unpublish the template sections today and put up a three-paragraph about page with the founder, the focus and three numbers.

    How to measure it

    About-to-contact rate

    Now
    –
    Target
    3% of about page visits
    By
    60 days

    Steps

    1. 1

      Take the template down

      Remove the Wonder hero, timeline, team and Lorem Ipsum from /about.

      Webflow Developer 1 hour

    2. 2

      Write the real story

      Founding year, AI-only focus, what ships, how a 31-day POC works.

      Founder & Chief Architect 1 day

    3. 3

      Add people and proof

      Real team roles and photos, partner badges, three numbered results.

      Marketing Lead 1 day

    4. 4

      Add schema

      Organization and founder markup with sameAs links.

      Webflow Developer 2 hours

    5. 5

      Link it

      Point the header About link at /about instead of "#".

      Webflow Developer 15 minutes

    6. 6

      Recrawl

      Request indexing in Search Console.

      Webflow Developer 15 minutes

    7. 7

      Correct third-party profiles

      Update company-data and directory listings that copied the template team.

      Marketing Lead 1 week

    First result
    1 week
    Cost
    $0, staff time only
    Owner
    Founder & Chief Architect

    Smallest test

    Replace the hero and timeline today and watch about-page exits for two weeks.

    Why it works

    Replace the template this week with a short, true about page: who founded Width and when, the team's AI-only focus, what the team ships (agentic systems, RAG, catalogue AI, n8n), the partner badges, three proof points with numbers (10,000+ users of a voice agent, a 14-agent system for a holding company serving 3,000 brands, 97% categorization accuracy), real photos and roles, and a "Book a scoping call" button. Add Organization schema with the founder and the LinkedIn profile, link the page from the header (today the About link points to "#"), and ask Google to recrawl it. Then correct any company-data profiles that copied the template leadership.

    Consulting buyers check the about page and the team before a first call, and a template page reads as a red flag on a site that sells careful engineering. The same page feeds the summaries AI assistants and data tools give about Width, so fixing it fixes the answers they give.

    Ready to use

    ChecklistAbout page replacement list

    [ ] Remove Wonder hero, timeline, team grid and Lorem Ipsum
    [ ] Founder, founding year and AI-only focus
    [ ] What we ship: agentic systems, RAG, catalogue AI, n8n
    [ ] Three numbered results with links to case studies
    [ ] Real team roles and photos
    [ ] Partner badges (AWS, Pinecone, n8n)
    [ ] Organization schema with founder and sameAs
    [ ] Header About link points to /about
    [ ] Request indexing
    [ ] Correct directory and data profiles

    Risks and how to handle them

    • The new page is thin

      Ship a short true page now and add team bios over the next month.

    • Old template text stays cached in AI answers

      Recrawl, add schema and correct the main directory profiles.

    Evidence

    • Your websiteAbout page (checked November 19, 2026): an unedited Webflow template for a company called "Wonder": title "Wonder - Webflow HTML Website Template", H1 "We're a team of creators and innovators", a timeline with "250,000 paying customer", "Wonder's IPO" and Lorem Ipsum, and a placeholder founder and 12-person team. Width appears only in the navigation and footer.Source
    • Company dataCompany data compiled from width.ai on November 19, 2026 lists the template's placeholder co-founders (a CFO and a CMO) as width.ai's leadership, with https://www.width.ai/about as the source.Source
    • Your websiteHomepage (checked November 19, 2026): H1 "Machine Learning, Ai, Natural Language Processing, Computer Vision, Width.ai"; 13 case study cards, mostly 2021-2023 projects such as "98.57% Accuracy Image Recognition in Restaurant SaaS", two listed twice; client logos with empty alt text; a testimonial about an unrelated all-in-one business tool shown three times; About, Blog and Contact header links pointing to "#"; and five blog preview "Read More" links that do not open width.ai posts.Source
    • Your websiteAgentic AI services page: "Your scoping call this week - POC in 31 days"; Hire-a-Dev, project-based and consulting models; proof missing from the homepage: a voice agent used by "10,000+ customers", a 14-node agent system for a holding company serving 3,000 brands, "96%+ field accuracy" for a medical-records startup and a 30% cut in sales team hours; visitors "leave the first call with a clear scope and the chatbot development cost"; no prices shown.Source
  • 02

    Width has 19 case studies, and the newest are strong: a 14-agent listing system for a holding company serving 3,000 brands (June 2026), 92.44% product similarity (July 2026), 89% retail-shelf SKU recognition (August 2026) and 92-97% categorization. The homepage shows 13 older cards, mostly first-generation language-model projects from 2021-2023, while the services page holds proof the homepage never mentions (10,000+ users of a voice agent, 96%+ field accuracy, 30% fewer sales hours).

    High impactLow effort First result in 2 weeks Marketing Lead

    First action

    Pick the six results and write a 12-word caption for each.

    How to measure it

    Homepage-to-contact conversion

    Now
    –
    Target
    +30%
    By
    60 days

    Steps

    1. 1

      Choose six results

      Newest studies plus the services-page proof.

      Founder & Chief Architect 1 hour

    2. 2

      Write captions

      Number, industry, what was built.

      Marketing Lead half a day

    3. 3

      Design the block

      Six cards plus a "what we build now" strip.

      Webflow Developer 1 day

    4. 4

      Archive the old cards

      Keep 2021-2023 studies on the case studies page under "Earlier work".

      Webflow Developer 2 hours

    5. 5

      Link each card

      To the full case study or service page.

      Webflow Developer 1 hour

    6. 6

      Measure

      Card clicks and contact rate in GA.

      Marketing Lead monthly

    First result
    2 weeks
    Cost
    $0, staff time only
    Owner
    Marketing Lead

    Smallest test

    Swap the first three cards for the three 2026 studies this week.

    Why it works

    Rebuild the homepage proof block around six current results, each with a number, an industry and a link: the agentic listing system, the voice agent, the medical-records extractor, categorization at 97%, SKU recognition at 89% and the 30% sales-hours cut. Move the 2021-2023 studies to an archive section on the case studies page. Add a one-line "what we build now" strip (agents, RAG, catalogue AI, n8n) above the cards.

    Buyers judge an AI firm by how current its work is. Showing agent and catalogue results from this year matches what they are shopping for and what the blog already ranks for.

    Risks and how to handle them

    • Client approval for named results

      Describe clients by industry where names are not cleared.

    Evidence

    • Your websiteCase studies page: 19 case studies; 13 date from 2021 to 2023. The newest are "Agentic AI for E-Commerce at Enterprise Scale" (June 22, 2026), a product-similarity study at 92.44% (July 16, 2026) and retail-shelf SKU recognition at 89% (August 30, 2026), plus three product-categorization studies at 92-97% accuracy (December 2024 to March 2025).Source
    • Your websiteHomepage (checked November 19, 2026): H1 "Machine Learning, Ai, Natural Language Processing, Computer Vision, Width.ai"; 13 case study cards, mostly 2021-2023 projects such as "98.57% Accuracy Image Recognition in Restaurant SaaS", two listed twice; client logos with empty alt text; a testimonial about an unrelated all-in-one business tool shown three times; About, Blog and Contact header links pointing to "#"; and five blog preview "Read More" links that do not open width.ai posts.Source
    • Your websiteAgentic AI services page: "Your scoping call this week - POC in 31 days"; Hire-a-Dev, project-based and consulting models; proof missing from the homepage: a voice agent used by "10,000+ customers", a 14-node agent system for a holding company serving 3,000 brands, "96%+ field accuracy" for a medical-records startup and a 30% cut in sales team hours; visitors "leave the first call with a clear scope and the chatbot development cost"; no prices shown.Source
  • 03

    The homepage H1, also its largest element, reads "Machine Learning, Ai, Natural Language Processing, Computer Vision, Width.ai", the same headline as a year ago. The strongest offer on the site, "Your scoping call this week - POC in 31 days", sits on a services page.

    Medium impactLow effort First result in 2 weeks Marketing Lead

    First action

    Write three H1 options and pick one with the founder.

    How to measure it

    Homepage bounce

    Now
    –
    Target
    -15%
    By
    45 days

    Steps

    1. 1

      Draft options

      Three H1s built on outcome and timeline.

      Marketing Lead 2 hours

    2. 2

      Pick one

      Founder chooses.

      Founder & Chief Architect 15 minutes

    3. 3

      Update hero

      H1, subline and one button.

      Webflow Developer 1 hour

    4. 4

      Update meta

      Title and description to match.

      Webflow Developer 30 minutes

    5. 5

      Keep keywords

      Service terms move to body copy and the services menu.

      Marketing Lead 1 hour

    6. 6

      Watch

      Bounce, scroll and Search Console queries.

      Marketing Lead monthly

    First result
    2 weeks
    Cost
    $0
    Owner
    Marketing Lead

    Smallest test

    Change the H1 and button today.

    Why it works

    Replace the H1 with a plain promise, for example "We build production AI agents and automations. Scoping call this week, working POC in 31 days." Follow it with one line on who it is for (ecommerce, marketplaces, services firms) and one primary button, "Book a scoping call". Update the title tag and meta description to match, and keep the service keywords in the body copy and navigation.

    A clear outcome and timeline in the first line tells a buyer in three seconds whether to stay, and the 31-day POC sets Width apart from rivals' generic "AI development company" headlines.

    Risks and how to handle them

    • Losing rankings for old keyword terms

      Keep the terms in headings lower on the page.

    Evidence

    • Your websiteHomepage (checked November 19, 2026): H1 "Machine Learning, Ai, Natural Language Processing, Computer Vision, Width.ai"; 13 case study cards, mostly 2021-2023 projects such as "98.57% Accuracy Image Recognition in Restaurant SaaS", two listed twice; client logos with empty alt text; a testimonial about an unrelated all-in-one business tool shown three times; About, Blog and Contact header links pointing to "#"; and five blog preview "Read More" links that do not open width.ai posts.Source
    • Your websiteAgentic AI services page: "Your scoping call this week - POC in 31 days"; Hire-a-Dev, project-based and consulting models; proof missing from the homepage: a voice agent used by "10,000+ customers", a 14-node agent system for a holding company serving 3,000 brands, "96%+ field accuracy" for a medical-records startup and a 30% cut in sales team hours; visitors "leave the first call with a clear scope and the chatbot development cost"; no prices shown.Source
    • Competitor researchRivals' own homepages: LeewayHertz, "AI development company enabling innovation and rapid development", with dedicated-team, team-extension and project models; Markovate, "300+ projects delivered", a scoped pilot of about 4-6 weeks and Forward Deployed Engineers; Addepto, "100+ AI projects" and managed delivery. None shows prices.Source
  • 04

    The homepage shows a testimonial praising an unrelated all-in-one tool for emailing clients and tracking expenses, three times. The About, Blog and Contact header links point to "#", two case studies appear twice, client logos (Coca-Cola, Munich Re, GM, Fitch Ratings and others) have empty alt text, and the five blog preview "Read More" links do not open width.ai posts, although the blog has 189 of its own.

    Medium impactLow effort First result in 1 week Webflow Developer

    First action

    Delete the stray testimonial and fix the three header links.

    How to measure it

    Homepage pages per session

    Now
    –
    Target
    +20%
    By
    30 days

    Steps

    1. 1

      Remove the stray testimonial

      All three copies.

      Webflow Developer 15 minutes

    2. 2

      Fix header links

      About, Blog and Contact to real pages.

      Webflow Developer 15 minutes

    3. 3

      De-duplicate cards

      Legal cover sheets and inventory automation.

      Webflow Developer 15 minutes

    4. 4

      Repoint blog cards

      Use a collection list of the latest width.ai posts.

      Webflow Developer 1 hour

    5. 5

      Add logo alt text

      Client name on each logo.

      Webflow Developer 30 minutes

    6. 6

      Schedule link checks

      Quarterly crawl for broken and off-site links.

      Marketing Lead quarterly

    First result
    1 week
    Cost
    $0
    Owner
    Webflow Developer

    Smallest test

    Do the first three steps in one sitting today.

    Why it works

    Delete the stray testimonial, point the header links at /about, /blog and /contact, remove the duplicate case study cards, add alt text naming each client logo, and repoint the blog preview cards at five current width.ai posts (or a live feed of the latest posts). Add a quarterly link check to catch the same problem.

    Small errors on a consultancy's homepage make buyers question how careful the engineering is. Working links also keep visitors, and their search value, on width.ai.

    Risks and how to handle them

    • Logo use not cleared

      Keep only logos with written permission.

    Evidence

    • Your websiteHomepage (checked November 19, 2026): H1 "Machine Learning, Ai, Natural Language Processing, Computer Vision, Width.ai"; 13 case study cards, mostly 2021-2023 projects such as "98.57% Accuracy Image Recognition in Restaurant SaaS", two listed twice; client logos with empty alt text; a testimonial about an unrelated all-in-one business tool shown three times; About, Blog and Contact header links pointing to "#"; and five blog preview "Read More" links that do not open width.ai posts.Source
    • Your websiteBlog (32 pages): recent posts on LangGraph, n8n, HubSpot-NetSuite integration, restaurant email and ecommerce; a 2023 resume-parser post is featured among 2026 posts; the only signup is the "Everything ML" newsletter ("Stay up to date with the world of machine learning & ai").Source
  • 05

    The contact page promises "Let's see if your project is a good fit" but asks only for name, email and a message. Width sells everything from free n8n setups to enterprise agent systems, so the founder cannot tell a $500 automation from a six-figure build until the call, and the Calendly scheduler is not tied to the answers.

    High impactLow effort First result in 2 weeks Founder & Chief Architect

    First action

    Write the three questions and their answer bands.

    How to measure it

    Call-to-proposal rate

    Now
    –
    Target
    +25%
    By
    60 days

    Steps

    1. 1

      Define bands

      Project types and budget ranges that match real deals.

      Founder & Chief Architect 1 hour

    2. 2

      Add fields

      Three dropdowns on /contact and /contact-us-rag.

      Webflow Developer 2 hours

    3. 3

      Route

      Qualified bands go to Calendly; small automation requests go to the package page.

      Webflow Developer half a day

    4. 4

      Tag in CRM

      Map fields to HubSpot properties.

      Marketing Lead 2 hours

    5. 5

      Promise a reply

      State a response time on the page.

      Founder & Chief Architect 15 minutes

    6. 6

      Review

      Monthly mix of leads by band and close rate.

      Founder & Chief Architect monthly

    First result
    2 weeks
    Cost
    $0
    Owner
    Founder & Chief Architect

    Smallest test

    Add only the project-type field this week.

    Why it works

    Add three quick fields: project type (agent or RAG system, catalogue AI or Pumice, n8n automation, hire a developer, advice), timeline and budget band. After submission, show Calendly right away for qualified bands, and send smaller automation requests to a fixed-price n8n package or the free workflows. Push the answers to HubSpot so each lead arrives tagged.

    Qualifying questions raise lead quality without hurting volume when they are multiple choice, and an instant booking for good-fit projects cuts days off the first call.

    Risks and how to handle them

    • Fewer submissions

      Make budget optional and compare volume after 30 days.

    Evidence

    • Your websiteContact page: "Let's see if your project is a good fit"; the form asks for first name, last name, email and a message only, with no project type, budget or timeline question and no response-time promise; a Calendly scheduler is also on the page.Source
    • Your website"How to Automate Restaurant Email Marketing: From someone who has done it 17 times" (November 19, 2026): free Square + Mailchimp workflows on GitHub, a "Get my free setup" button and "You only pay if you find value"; no restaurant client is named.Source
    • Your websiteAgentic AI services page: "Your scoping call this week - POC in 31 days"; Hire-a-Dev, project-based and consulting models; proof missing from the homepage: a voice agent used by "10,000+ customers", a 14-node agent system for a holding company serving 3,000 brands, "96%+ field accuracy" for a medical-records startup and a 30% cut in sales team hours; visitors "leave the first call with a clear scope and the chatbot development cost"; no prices shown.Source
  • 06

    Width gives away working n8n workflows (Google review replies, restaurant email) as GitHub downloads with no email capture, so it never learns who installed them. The review-replies post has an empty "Keep Reading" section and links no case study, although "5 n8n Marketing Automation Use Cases We Run for Real Customers" exists.

    Medium impactLow effort First result in 3 weeks N8n Automation Engineer

    First action

    Add the optional email box to the Google reviews post.

    How to measure it

    Workflow-post email capture

    Now
    –
    Target
    5% of readers
    By
    60 days

    Steps

    1. 1

      Add the box

      Optional email for setup guide and updates.

      Webflow Developer 2 hours

    2. 2

      Write the sequence

      Three emails: setup, use case, custom build offer.

      n8n Automation Engineer 1 day

    3. 3

      Tag by workflow

      HubSpot property per download.

      Marketing Lead 1 hour

    4. 4

      Fill Keep Reading

      Use-case post and one case study on each post.

      Webflow Developer 2 hours

    5. 5

      Track installs

      Add a UTM to the GitHub link and watch repo stars.

      n8n Automation Engineer 1 hour

    6. 6

      Review

      Which workflows produce calls.

      Marketing Lead monthly

    First result
    3 weeks
    Cost
    $0
    Owner
    N8n Automation Engineer

    Smallest test

    Run the box on one post for a month.

    Why it works

    Keep the GitHub link free, and add an optional "Email me the setup guide and updates" box next to it, plus a short follow-up sequence: setup tips, one customer use case, then an offer of a fixed-price custom build. Fill "Keep Reading" on every workflow post with the n8n use-case post and one relevant case study, and tag downloads by workflow in HubSpot.

    People who install an automation have the problem and the tools, which makes them the warmest leads Width has, and a soft capture keeps the giveaway generous.

    Risks and how to handle them

    • Looks like a bait-and-switch

      Keep the download ungated; the email box is optional.

    Evidence

    • Your website"Automate Google Reviews With AI Replies (Free n8n Workflow)" (November 19, 2026): the workflow downloads from GitHub with no email capture; CTAs offer free setup and "Talk to us about your reviews"; the "Keep Reading" section is empty and no case study is linked.Source
    • Your websiten8n tag page: seven posts, including "5 n8n Marketing Automation Use Cases We Run for Real Customers" and "n8n vs LangGraph for Production AI Workflows", with no n8n service offer, only "Get In Touch" and a newsletter signup.Source
    • Your website"How to Automate Restaurant Email Marketing: From someone who has done it 17 times" (November 19, 2026): free Square + Mailchimp workflows on GitHub, a "Get my free setup" button and "You only pay if you find value"; no restaurant client is named.Source
  • 07

    width.ai/robots.txt holds only "Allow: /" and a sitemap line with no User-agent line, and Lighthouse reports "robots.txt is not valid". The sitemap still lists Webflow template pages (/template-info/style-guide with its Lorem ipsum, /changelog, /licenses), /blog-old and two Heatrr.ai pages, next to 189 real posts.

    Medium impactLow effort First result in 2 weeks Webflow Developer

    First action

    Paste a valid robots.txt into Webflow's SEO settings.

    How to measure it

    Junk pages indexed

    Now
    –
    Target
    0
    By
    30 days

    Steps

    1. 1

      Fix robots.txt

      "User-agent: *", "Allow: /" and the sitemap URL.

      Webflow Developer 15 minutes

    2. 2

      List junk pages

      Template-info, blog-old, test and success pages.

      Webflow Developer 1 hour

    3. 3

      Noindex or delete

      Exclude them from the sitemap.

      Webflow Developer 1 hour

    4. 4

      Redirect

      301s for anything with backlinks.

      Webflow Developer 1 hour

    5. 5

      Resubmit

      Sitemap in Search Console.

      Webflow Developer 15 minutes

    6. 6

      Check coverage

      Indexed pages report after two weeks.

      Marketing Lead 2 weeks

    First result
    2 weeks
    Cost
    $0
    Owner
    Webflow Developer

    Smallest test

    Fix robots.txt today and rerun Lighthouse.

    Why it works

    Rewrite robots.txt with a User-agent line and the sitemap. Set the template-info pages and /blog-old to noindex and remove them from the sitemap (or delete them), redirect anything with links to the closest live page, and resubmit the sitemap in Search Console.

    A valid robots file and a clean sitemap let search engines and AI crawlers spend their time on the posts and service pages that bring in buyers, not on template filler.

    Risks and how to handle them

    • Noindexing a page that ranks

      Check Search Console clicks before removing anything.

    Evidence

    • Your websiterobots.txt holds only "Allow: /" and the sitemap line, with no User-agent line, and Lighthouse reports "robots.txt is not valid"; /template-info/style-guide still shows the template's Lorem ipsum style guide.Source
    • Your websiteSitemap (266 URLs, checked November 19, 2026): 189 blog posts; service pages for chatbots, RAG, agentic AI, prompt engineers, NLP and RAG for healthcare; ebook pages for agentic RAG, AI system evals, content automation, AI for ecommerce and an Upwork playbook; no n8n service page and no post on project cost; Webflow template pages (/template-info/style-guide, /changelog, /licenses), /blog-old and two Heatrr.ai pages still listed.Source
    • PageSpeedAccessibility 84 and SEO 83: no lang attribute on the <html> element, low text contrast, links without a discernible name or descriptive text, and small touch targets.Source
  • 08

    width.ai/heatrr-ai sells heatmap optimization, but its headline reads "Use Ai To Generate Hyper Personalized Cold Emails That, Heatrr.ai". Its testimonials name other companies, it promises "Completely Risk Free - 100% Money Back", and it has no dates or prices. A second Heatrr page is also in the sitemap.

    Low impactLow effort First result in 1 week Marketing Lead

    First action

    Ask the founder whether Heatrr is still sold.

    How to measure it

    Broken side pages

    Now
    –
    Target
    0
    By
    30 days

    Steps

    1. 1

      Decide

      Keep, merge or retire Heatrr.

      Founder & Chief Architect 15 minutes

    2. 2

      Check traffic

      Search Console clicks and backlinks for both pages.

      Marketing Lead 1 hour

    3. 3

      Redirect or rewrite

      301 to the best live page, or fix headline and proof.

      Webflow Developer 2 hours

    4. 4

      Check the guarantee

      Remove the money-back line unless it is honoured.

      Founder & Chief Architect 15 minutes

    5. 5

      Sweep other side pages

      Same check on name-matching and product-matching API pages.

      Marketing Lead half a day

    6. 6

      Update sitemap

      Remove retired URLs.

      Webflow Developer 15 minutes

    First result
    1 week
    Cost
    $0
    Owner
    Marketing Lead

    Smallest test

    Fix the headline today while deciding the rest.

    Why it works

    Decide whether Heatrr is still offered. If not, 301 both pages to the most relevant live service page (ecommerce AI or analytics). If it is, fix the headline, replace the testimonials with real Heatrr results, and check that the money-back promise is one Width honours.

    Side pages with copy errors and borrowed testimonials weaken trust in the main brand, and redirecting them passes their search value to pages that sell.

    Risks and how to handle them

    • A few Heatrr enquiries are lost

      Redirect to a page that offers the same analytics help.

    Evidence

    • Your websiteHeatrr.ai page on width.ai: a heatmap-optimization service whose headline reads "Use Ai To Generate Hyper Personalized Cold Emails That, Heatrr.ai"; its testimonials name other companies; "Completely Risk Free - 100% Money Back"; no dates or prices.Source
    • Your websiteSitemap (266 URLs, checked November 19, 2026): 189 blog posts; service pages for chatbots, RAG, agentic AI, prompt engineers, NLP and RAG for healthcare; ebook pages for agentic RAG, AI system evals, content automation, AI for ecommerce and an Upwork playbook; no n8n service page and no post on project cost; Webflow template pages (/template-info/style-guide, /changelog, /licenses), /blog-old and two Heatrr.ai pages still listed.Source
  • 09

    Google Lighthouse scores the homepage 84 for accessibility: the <html> element has no lang attribute, text contrast is too low in places, some links have no discernible name or descriptive text, touch targets are too small, and there is no main landmark. Enterprise buyers increasingly check vendor sites against accessibility standards.

    Low impactLow effort First result in 1 week Webflow Developer

    First action

    Add lang="en" in Webflow's site settings.

    How to measure it

    Accessibility score

    Now
    –
    Target
    95+
    By
    14 days

    Steps

    1. 1

      Set language

      lang="en" on html.

      Webflow Developer 10 minutes

    2. 2

      Fix contrast

      Darken the failing text colours.

      Webflow Developer 1 hour

    3. 3

      Name links

      Descriptive text or aria-labels.

      Webflow Developer 2 hours

    4. 4

      Enlarge tap targets

      At least 44px on mobile.

      Webflow Developer 1 hour

    5. 5

      Add a main landmark

      Wrap page content.

      Webflow Developer 30 minutes

    6. 6

      Retest

      Lighthouse on the homepage and two service pages.

      Webflow Developer 30 minutes

    First result
    1 week
    Cost
    $0
    Owner
    Webflow Developer

    Smallest test

    Fix lang and link names first and retest.

    Why it works

    Set lang="en" on the site, raise contrast on the failing text, give icon and image links accessible names ("Read the IPG case study" rather than "Learn more"), enlarge mobile tap targets, and wrap the content in a main landmark. Rerun Lighthouse and keep the score above 95.

    These are one-day fixes that also help SEO (descriptive link text and language) and remove an easy objection in enterprise vendor reviews.

    Risks and how to handle them

    • Design changes

      Adjust shades, not the palette.

    Evidence

    • PageSpeedAccessibility 84 and SEO 83: no lang attribute on the <html> element, low text contrast, links without a discernible name or descriptive text, and small touch targets.Source

This quarter 11

The core plays for the next three months.

  • 10

    The homepage carries an n8n Expert Partner badge and a "Top Ranked n8n Agency" badge, and the blog publishes n8n tutorials, comparisons and "5 n8n Marketing Automation Use Cases We Run for Real Customers". Yet the services menu lists computer vision, chatbots, object detection, prompt engineering, NLP and data visualization, the sitemap has no n8n service page, and the n8n tag page ends with "Get In Touch".

    High impactMedium effort First result in 1 month N8n Automation Engineer

    First action

    Draft the three packages with a starting price for each.

    How to measure it

    n8n enquiries

    Now
    –
    Target
    10 a month
    By
    90 days

    Steps

    1. 1

      Define packages

      Audit, fixed workflow, care plan.

      n8n Automation Engineer 1 day

    2. 2

      Gather proof

      Use cases, HubSpot-NetSuite build, review workflow.

      n8n Automation Engineer 2 days

    3. 3

      Write the page

      Offer, packages, proof, FAQ, n8n vs LangGraph.

      Content Writer 3 days

    4. 4

      Build

      Page plus services menu entry.

      Webflow Developer 1 day

    5. 5

      Link

      From all n8n posts and the n8n tag page.

      Webflow Developer 2 hours

    6. 6

      List it

      Use the page on partner directory profiles.

      Marketing Lead 1 day

    7. 7

      Measure

      Enquiries tagged n8n.

      Marketing Lead monthly

    First result
    1 month
    Cost
    Staff time only
    Owner
    N8n Automation Engineer

    Smallest test

    Publish a one-screen version with one package and a booking link.

    Why it works

    Publish /n8n-automation-services: what Width builds in n8n (AI agents, CRM and ERP sync such as HubSpot-NetSuite, marketing and review automations), three packages (audit and fix, a fixed-scope workflow, ongoing care), proof from the use-case post, the partner badges, and how n8n and LangGraph fit together for production. Add it to the services menu and link it from every n8n post.

    Readers of n8n tutorials are often a week away from hiring help. A named offer turns that traffic into enquiries, and the page can rank for "n8n agency" searches, which rivals target with dedicated pages.

    Risks and how to handle them

    • Small, low-budget n8n jobs

      Set package minimums and route very small jobs to the free workflows.

    Evidence

    • Your websiten8n tag page: seven posts, including "5 n8n Marketing Automation Use Cases We Run for Real Customers" and "n8n vs LangGraph for Production AI Workflows", with no n8n service offer, only "Get In Touch" and a newsletter signup.Source
    • Your websiteHomepage proof: testimonials from WorkOtter, Denison University, SalesHawk and JDM Garages ("Width.ai increased our advertiser revenue 36% in just that first month"); "over 50 services companies have worked with us"; badges for n8n Expert Partner, "Top Ranked n8n Agency", AWS Partner Advanced Tier, Azure AI Certified and Pinecone Partner; the "Everything ML" newsletter.Source
    • Your websiteSitemap (266 URLs, checked November 19, 2026): 189 blog posts; service pages for chatbots, RAG, agentic AI, prompt engineers, NLP and RAG for healthcare; ebook pages for agentic RAG, AI system evals, content automation, AI for ecommerce and an Upwork playbook; no n8n service page and no post on project cost; Webflow template pages (/template-info/style-guide, /changelog, /licenses), /blog-old and two Heatrr.ai pages still listed.Source
  • 11

    Width's best offer is "Your scoping call this week - POC in 31 days", and it promises visitors "leave the first call with a clear scope and the chatbot development cost". No price appears anywhere. LeewayHertz, Markovate (a 4-6 week pilot) and Addepto don't publish prices either, so buyers still can't compare.

    High impactMedium effort First result in 1 month Founder & Chief Architect

    First action

    Pull the scope and price of the last ten POCs to set the band.

    How to measure it

    POCs sold

    Now
    –
    Target
    +50%
    By
    2 quarters

    Steps

    1. 1

      Set the band

      From the last ten POCs.

      Founder & Chief Architect 1 day

    2. 2

      Define included work

      Deliverables, eval report, production plan.

      Founder & Chief Architect 1 day

    3. 3

      Write the page

      Band, inclusions, timeline, FAQ.

      Content Writer 2 days

    4. 4

      Build and link

      Hero, services pages and contact form.

      Webflow Developer 1 day

    5. 5

      Add a starter tier

      Fixed price for one well-defined use case.

      Founder & Chief Architect 1 week

    6. 6

      Track

      Leads, calls and POCs closed against before.

      Marketing Lead monthly

    First result
    1 month
    Cost
    Staff time only
    Owner
    Founder & Chief Architect

    Smallest test

    Add one line, "POCs typically range from $X to $Y", to the agentic services page.

    Why it works

    Create a /31-day-poc page with what is included (scoping, a working build on the buyer's data, an eval report, a production plan), what is not, a "POCs typically range from $X to $Y" band based on the last ten projects, and a fixed-price starter tier for well-defined use cases. Link it from the homepage hero, the services pages and the contact form's budget field.

    A visible price band removes the biggest unknown for a first AI project, filters out mismatched leads, and in a market where nobody shows prices it makes Width the easy one to shortlist.

    Risks and how to handle them

    • Anchoring too low

      Publish a range with what drives the top end.

    • Rivals copy it

      Being first and specific still wins searches and shortlists.

    Builds on

    Evidence

    • Your websiteAgentic AI services page: "Your scoping call this week - POC in 31 days"; Hire-a-Dev, project-based and consulting models; proof missing from the homepage: a voice agent used by "10,000+ customers", a 14-node agent system for a holding company serving 3,000 brands, "96%+ field accuracy" for a medical-records startup and a 30% cut in sales team hours; visitors "leave the first call with a clear scope and the chatbot development cost"; no prices shown.Source
    • Competitor researchRivals' own homepages: LeewayHertz, "AI development company enabling innovation and rapid development", with dedicated-team, team-extension and project models; Markovate, "300+ projects delivered", a scoped pilot of about 4-6 weeks and Forward Deployed Engineers; Addepto, "100+ AI projects" and managed delivery. None shows prices.Source
  • 12

    Width's flagship 2026 case study, a 14-agent Amazon listing system for Interpublic Group across 3,000 brands, reports no measured result: no hours saved, no accuracy and no sales lift, only that it replaces "days of manual research". The older studies lead with numbers (98.57%, 167.3 hr/mo, 36%), and the newest, biggest one does not.

    Medium impactMedium effort First result in 1 month Founder & Chief Architect

    First action

    Email the client contact asking for before-and-after time per brief.

    How to measure it

    Measured case studies

    Now
    –
    Target
    All 2025-2026 studies
    By
    1 quarter

    Steps

    1. 1

      Ask for data

      Time per brief, volume, listing outcomes.

      Founder & Chief Architect 1 week

    2. 2

      Compute results

      Hours saved and throughput.

      Lead AI Engineer 2 days

    3. 3

      Clear wording

      Client approval for numbers and a quote.

      Founder & Chief Architect 2 weeks

    4. 4

      Design the box

      Standard results box for all studies.

      Webflow Developer 1 day

    5. 5

      Update studies

      IPG first, then the 2026 SKU and product-similarity studies.

      Content Writer 1 week

    6. 6

      Reuse

      Numbers on the homepage, about and service pages.

      Marketing Lead 1 day

    First result
    1 month
    Cost
    Staff time only
    Owner
    Founder & Chief Architect

    Smallest test

    Add a time-per-brief figure to the IPG study.

    Why it works

    Work with the client to measure three things: time per listing brief before and after, briefs produced per strategist per week, and, where allowed, conversion or ranking change on optimized listings. Add a results box at the top of the IPG study and make a standard box (problem, numbers, stack, timeline) for every new case study. Ask for a quotable line from the client team.

    Buyers forward numbers internally, not architecture diagrams. A measured result on an enterprise agent system is the strongest proof Width can show for its main offer.

    Risks and how to handle them

    • Client will not share numbers

      Report internal benchmark numbers from the build, labelled as such.

    Evidence

    • Your website"Agentic AI for E-Commerce at Enterprise Scale" (June 22, 2026): an Amazon listing system for Interpublic Group with "roughly fourteen ReAct agent nodes" across 3,000 brands; it reports no measured result (time saved, accuracy or sales lift), only that it replaces "days of manual research".Source
    • Your websiteCase studies page: 19 case studies; 13 date from 2021 to 2023. The newest are "Agentic AI for E-Commerce at Enterprise Scale" (June 22, 2026), a product-similarity study at 92.44% (July 16, 2026) and retail-shelf SKU recognition at 89% (August 30, 2026), plus three product-categorization studies at 92-97% accuracy (December 2024 to March 2025).Source
  • 13

    Width publishes about a post a day on agents, RAG and n8n, 189 posts in all, but none on cost, the question every buyer asks first. The services page already promises the cost on the first call, and none of the rivals checked publishes prices.

    High impactMedium effort First result in 6 weeks Content Writer

    First action

    List cost drivers and ranges from the last 20 projects.

    How to measure it

    Cost-guide leads

    Now
    –
    Target
    15 a month
    By
    90 days

    Steps

    1. 1

      Gather data

      Ranges and drivers from past projects.

      Founder & Chief Architect 1 day

    2. 2

      Outline

      Project types, drivers, running costs.

      Content Writer 1 day

    3. 3

      Write

      With worked examples.

      Content Writer 1 week

    4. 4

      Build the estimator

      Three questions that give a range and prefill the form.

      Webflow Developer 3 days

    5. 5

      Link

      From service pages and related posts.

      Webflow Developer 2 hours

    6. 6

      Refresh

      Update ranges twice a year.

      Content Writer 6 months

    First result
    6 weeks
    Cost
    Staff time only
    Owner
    Content Writer

    Smallest test

    Publish the chatbot section first.

    Why it works

    Write a cost guide from Width's own project data: typical ranges for a chatbot, a RAG system, an agentic workflow, catalogue AI and an n8n automation; what drives cost (integrations, data quality, evals, real-time APIs); build versus hire versus Pumice; and running costs after launch. Add a simple estimator that feeds the contact form, and link it from every service page.

    Cost searches come from buyers with a budget to set, and an honest guide becomes the page AI assistants and buyers quote, with Width as the source.

    Risks and how to handle them

    • Ranges quoted out of context

      Show the drivers next to every range.

    Builds on

    Evidence

    • Your websiteSitemap (266 URLs, checked November 19, 2026): 189 blog posts; service pages for chatbots, RAG, agentic AI, prompt engineers, NLP and RAG for healthcare; ebook pages for agentic RAG, AI system evals, content automation, AI for ecommerce and an Upwork playbook; no n8n service page and no post on project cost; Webflow template pages (/template-info/style-guide, /changelog, /licenses), /blog-old and two Heatrr.ai pages still listed.Source
    • Your websiteAgentic AI services page: "Your scoping call this week - POC in 31 days"; Hire-a-Dev, project-based and consulting models; proof missing from the homepage: a voice agent used by "10,000+ customers", a 14-node agent system for a holding company serving 3,000 brands, "96%+ field accuracy" for a medical-records startup and a 30% cut in sales team hours; visitors "leave the first call with a clear scope and the chatbot development cost"; no prices shown.Source
    • Competitor researchRivals' own homepages: LeewayHertz, "AI development company enabling innovation and rapid development", with dedicated-team, team-extension and project models; Markovate, "300+ projects delivered", a scoped pilot of about 4-6 weeks and Forward Deployed Engineers; Addepto, "100+ AI projects" and managed delivery. None shows prices.Source
  • 14

    Width's strongest recent proof is catalogue work: categorization at 92-97% accuracy (one with 5,585 categories), 89% retail-shelf SKU recognition and 92.44% product similarity. Pumice.ai sells the same capability as "Ai Agents for SKU Onboarding & Product Data Enrichment" but names no customers, and width.ai's menu links to Pumice without explaining that they are one team.

    High impactMedium effort First result in 6 weeks Pumice Product Lead

    First action

    List every catalogue-related page, post and case study across both sites.

    How to measure it

    Catalogue AI leads

    Now
    –
    Target
    +50%
    By
    2 quarters

    Steps

    1. 1

      Inventory pages

      Case studies, posts, API pages, ebook.

      Content Writer 1 day

    2. 2

      Write the hub

      Problems, proof, product or custom build.

      Content Writer 1 week

    3. 3

      Build it

      Hub page and menu link.

      Webflow Developer 2 days

    4. 4

      Cross-link Pumice

      Case studies on pumice.ai; hub link back.

      Pumice Product Lead 1 day

    5. 5

      Redirect thin pages

      Fold overlapping API pages into the hub where they duplicate it.

      Webflow Developer half a day

    6. 6

      Route leads

      One form that tags software or custom build.

      Marketing Lead half a day

    7. 7

      Measure

      Leads and demos by source.

      Marketing Lead monthly

    First result
    6 weeks
    Cost
    Staff time only
    Owner
    Pumice Product Lead

    Smallest test

    Add the three categorization case studies to pumice.ai this week.

    Why it works

    Create /ecommerce-catalog-ai on width.ai: the problems (supplier onboarding, taxonomy, attributes, matching, channel copy), the case studies with their numbers, when to use Pumice and when to commission a custom build, and one form. On pumice.ai, add a "Built by the team behind" section linking those case studies. Point the AI for ecommerce ebook, the product-matching API page and the related posts at the hub.

    Catalogue teams buy on accuracy evidence. Putting the evidence next to the product, with a clear choice between software and a custom build, converts both kinds of buyer.

    Risks and how to handle them

    • Confusing two brands

      State clearly that Pumice is the software and Width the team that builds custom systems.

    Evidence

    • Your websiteCase studies page: 19 case studies; 13 date from 2021 to 2023. The newest are "Agentic AI for E-Commerce at Enterprise Scale" (June 22, 2026), a product-similarity study at 92.44% (July 16, 2026) and retail-shelf SKU recognition at 89% (August 30, 2026), plus three product-categorization studies at 92-97% accuracy (December 2024 to March 2025).Source
    • Pumice.aiPumice.ai, "Powered by Width.ai": "Ai Agents for SKU Onboarding & Product Data Enrichment", categorization at "over 97% accuracy", catalog PDFs to CSV and channel copy for Amazon, Walmart and eBay; no named customers; main CTA "Request a Demo".Source
    • Your websiteSitemap (266 URLs, checked November 19, 2026): 189 blog posts; service pages for chatbots, RAG, agentic AI, prompt engineers, NLP and RAG for healthcare; ebook pages for agentic RAG, AI system evals, content automation, AI for ecommerce and an Upwork playbook; no n8n service page and no post on project cost; Webflow template pages (/template-info/style-guide, /changelog, /licenses), /blog-old and two Heatrr.ai pages still listed.Source
  • 15

    Pumice's pricing is custom per endpoint and per record, with no listed prices and no free trial, and every path ends at "Request a Demo". Its main claim, categorization at "over 97% accuracy", is something a buyer can check on their own data in minutes.

    High impactMedium effort First result in 1 month Pumice Product Lead

    First action

    Define the CSV template and the 24-hour manual process.

    How to measure it

    Sample-to-paid rate

    Now
    –
    Target
    20%
    By
    2 quarters

    Steps

    1. 1

      Template

      CSV columns and taxonomy upload.

      Pumice Product Lead 1 day

    2. 2

      Form

      Upload page on pumice.ai.

      Webflow Developer 2 days

    3. 3

      Run manually

      Team runs the endpoints and checks output.

      Lead AI Engineer ongoing

    4. 4

      Results email

      File, accuracy summary, volume price.

      Pumice Product Lead 2 days

    5. 5

      Follow up

      Call within two days of results.

      Founder & Chief Architect ongoing

    6. 6

      Automate

      Self-serve once volume justifies it.

      Lead AI Engineer 1 quarter

    First result
    1 month
    Cost
    Compute and team time per sample
    Owner
    Pumice Product Lead

    Smallest test

    Offer the sample run to the next five demo requests.

    Why it works

    Add a "Try it on 100 SKUs" flow: the buyer uploads a CSV and their taxonomy (or picks Google Product Taxonomy), and receives categories, attributes and a title rewrite with confidence scores by email within 24 hours, run by the team at first. The follow-up shows the accuracy on their file and a price for their volume.

    Catalogue managers trust their own data over a demo. A sample run proves the 97% claim on the buyer's own SKUs and gives sales a concrete reason to follow up.

    Risks and how to handle them

    • Messy files lower accuracy

      Show accuracy with a short data-quality note.

    • Too many tyre-kickers

      Require a work email and catalogue size.

    Builds on

    Evidence

    • Pumice.aiPumice pricing page: "We price based on the number of our endpoints you need available, with per record costs for each endpoint"; no listed prices or free trial; a default monthly credit allotment.Source
    • Pumice.aiPumice.ai, "Powered by Width.ai": "Ai Agents for SKU Onboarding & Product Data Enrichment", categorization at "over 97% accuracy", catalog PDFs to CSV and channel copy for Amazon, Walmart and eBay; no named customers; main CTA "Request a Demo".Source
  • 16

    Width's newest posts target local businesses: restaurant email marketing ("From someone who has done it 17 times") and Google review replies, each with free workflows, a free setup offer and "You only pay if you find value". Free setups generate goodwill but no defined revenue, and no restaurant client is named.

    Medium impactMedium effort First result in 1 month N8n Automation Engineer

    First action

    Price the bundle from the time the last free setups took.

    How to measure it

    Local packages sold

    Now
    –
    Target
    8 a month
    By
    90 days

    Steps

    1. 1

      Time past setups

      Hours per setup and support.

      n8n Automation Engineer 1 day

    2. 2

      Price

      Setup fee and care plan.

      Founder & Chief Architect 1 day

    3. 3

      Page

      One page with checkout or booking.

      Webflow Developer 3 days

    4. 4

      Case studies

      Two restaurant results with permission.

      Content Writer 2 weeks

    5. 5

      Swap CTAs

      Free-setup buttons offer free DIY or the paid package.

      Webflow Developer 2 hours

    6. 6

      Review

      Sales, hours and churn monthly.

      Founder & Chief Architect monthly

    First result
    1 month
    Cost
    Staff time only
    Owner
    N8n Automation Engineer

    Smallest test

    Offer the package to the next ten free-setup requests.

    Why it works

    Package the workflows: a one-time setup fee for reviews plus email plus a monthly report, with an optional monthly care plan, sold on one page with a clear price. Keep the free GitHub versions for people who set it up themselves. Turn the 17 restaurant setups into two named case studies with before-and-after numbers.

    Local businesses buy simple, priced bundles, and Width already has the workflows and the content traffic, so a price turns that work into repeatable revenue.

    Risks and how to handle them

    • Low-value support load

      Limit the care plan to defined changes per month.

    Builds on

    Evidence

    • Your website"How to Automate Restaurant Email Marketing: From someone who has done it 17 times" (November 19, 2026): free Square + Mailchimp workflows on GitHub, a "Get my free setup" button and "You only pay if you find value"; no restaurant client is named.Source
    • Your website"Automate Google Reviews With AI Replies (Free n8n Workflow)" (November 19, 2026): the workflow downloads from GitHub with no email capture; CTAs offer free setup and "Talk to us about your reviews"; the "Keep Reading" section is empty and no case study is linked.Source
  • 17

    width.ai hosts "The exact playbook we use to land 3-5 $100/hr Upwork clients per month", aimed at freelancers including beginners, alongside enterprise agent and RAG offers. Those signups land in the same lists as buyers, and a prospect who finds the page sees Width presenting itself as a $100/hr Upwork shop.

    Medium impactLow effort First result in 2 weeks Marketing Lead

    First action

    Count how many contacts came in through the Upwork playbook.

    How to measure it

    Buyer list quality

    Now
    –
    Target
    Freelancer signups excluded 100%
    By
    30 days

    Steps

    1. 1

      Count

      Contacts from the playbook form.

      Marketing Lead 1 hour

    2. 2

      Decide

      Move or segment.

      Founder & Chief Architect 15 minutes

    3. 3

      Tag

      Separate HubSpot audience.

      Marketing Lead 1 hour

    4. 4

      Exclude

      From sales sequences and the buyer newsletter.

      Marketing Lead 1 hour

    5. 5

      Move or hide

      Redirect, or remove from the sitemap and navigation.

      Webflow Developer 1 hour

    6. 6

      Check reports

      Conversion rates without freelancers.

      Marketing Lead monthly

    First result
    2 weeks
    Cost
    $0
    Owner
    Marketing Lead

    Smallest test

    Tag and exclude the freelancer signups this week.

    Why it works

    Either move the playbook to the founder's personal site or a separate brand, with a 301 redirect, or keep it but tag those signups as a separate audience in HubSpot, exclude them from sales sequences and the buyer newsletter, and remove the page from the main navigation and sitemap.

    Clean lists give honest conversion numbers and keep buyer emails relevant, and separating audiences protects premium positioning without losing the freelancer audience.

    Risks and how to handle them

    • Losing a referral source

      Keep a link from the founder's personal channels.

    Evidence

    • Your websiteUpwork playbook ebook page: "The exact playbook we use to land 3-5 $100/hr Upwork clients per month", aimed at freelancers including beginners; it collects name and email on the same domain as the consulting offer.Source
    • Your websiteSitemap (266 URLs, checked November 19, 2026): 189 blog posts; service pages for chatbots, RAG, agentic AI, prompt engineers, NLP and RAG for healthcare; ebook pages for agentic RAG, AI system evals, content automation, AI for ecommerce and an Upwork playbook; no n8n service page and no post on project cost; Webflow template pages (/template-info/style-guide, /changelog, /licenses), /blog-old and two Heatrr.ai pages still listed.Source
  • 18

    The only signup on the blog is "Everything ML" ("Stay up to date with the world of machine learning & ai"), while Width publishes in clear clusters (n8n, agents and RAG, ecommerce catalogue, local business) and has five ebooks. A general ML newsletter competes with hundreds of others and tells Width nothing about what a subscriber wants to buy.

    Medium impactMedium effort First result in 6 weeks Marketing Lead

    First action

    Count subscribers and map ebook signups to the three tracks.

    How to measure it

    Calls from newsletter

    Now
    –
    Target
    5 a month
    By
    2 quarters

    Steps

    1. 1

      Map list

      Ebook and workflow signups to tracks.

      Marketing Lead 1 day

    2. 2

      Set up tracks

      HubSpot lists and signup choice.

      Marketing Lead 1 day

    3. 3

      Template

      Post, workflow, result, offer.

      Content Writer 1 day

    4. 4

      Picker email

      Ask existing subscribers to choose.

      Marketing Lead 1 day

    5. 5

      Send monthly

      Three short issues.

      Content Writer monthly

    6. 6

      Score

      Replies and clicks feed lead scores.

      Marketing Lead monthly

    First result
    6 weeks
    Cost
    Staff time only
    Owner
    Marketing Lead

    Smallest test

    Send one n8n-track issue to n8n workflow downloaders.

    Why it works

    Relaunch as three short monthly tracks, chosen at signup and inferred from which ebook or workflow a contact downloaded. Each issue has one new post, one working workflow or code sample, one case result and one offer that fits the track (POC, sample run, n8n package). Send past ebook signups a one-click track picker.

    Topic tracks raise opens and replies, and the track itself is a buying signal sales can act on.

    Risks and how to handle them

    • Three newsletters is too much work

      Reuse the week's posts; keep each issue under 200 words.

    Builds on

    Evidence

    • Your websiteBlog (32 pages): recent posts on LangGraph, n8n, HubSpot-NetSuite integration, restaurant email and ecommerce; a 2023 resume-parser post is featured among 2026 posts; the only signup is the "Everything ML" newsletter ("Stay up to date with the world of machine learning & ai").Source
    • Your websiteSitemap (266 URLs, checked November 19, 2026): 189 blog posts; service pages for chatbots, RAG, agentic AI, prompt engineers, NLP and RAG for healthcare; ebook pages for agentic RAG, AI system evals, content automation, AI for ecommerce and an Upwork playbook; no n8n service page and no post on project cost; Webflow template pages (/template-info/style-guide, /changelog, /licenses), /blog-old and two Heatrr.ai pages still listed.Source
  • 19

    Real visitors pass Core Web Vitals (field LCP 1.5 s), so speed is not urgent. But the homepage downloads 3.8 MB in 115 requests, 3.2 MB of it from 12 third-party hosts, including several overlapping marketing tools (Hotjar, a HubSpot form connector, OptinMonster and Leadfeeder alongside Google Tag Manager). Render-blocking requests cost about 1 s, images could shrink by 655 KB, and 628 KB of JavaScript goes unused.

    Low impactLow effort First result in 2 weeks Webflow Developer

    First action

    List every third-party tool with an owner and a reason to keep it.

    How to measure it

    Homepage weight

    Now
    –
    Target
    Under 2 MB
    By
    30 days

    Steps

    1. 1

      Inventory tools

      Owner and purpose for each.

      Marketing Lead 2 hours

    2. 2

      Remove unused

      Drop overlapping tools.

      Webflow Developer 1 hour

    3. 3

      Defer the rest

      Load after render through Tag Manager.

      Webflow Developer 2 hours

    4. 4

      Compress images

      WebP and correct sizes.

      Webflow Developer 2 hours

    5. 5

      Fonts

      Preload one, defer others.

      Webflow Developer 1 hour

    6. 6

      Set a budget

      New tools need an owner and a weight check.

      Marketing Lead ongoing

    First result
    2 weeks
    Cost
    $0
    Owner
    Webflow Developer

    Smallest test

    Compress homepage images and recheck PageSpeed.

    Why it works

    List each tool, its owner and what it is used for; remove the ones nobody uses, and load the rest through Tag Manager after the page renders. Compress and resize the homepage images (mostly logos and case study cards), defer non-critical fonts and scripts, and set a page-weight budget for new tools.

    A lighter page keeps mobile visitors on slow connections and makes it clear who owns each tool, without chasing scores real users already pass.

    Risks and how to handle them

    • Losing tracking data

      Confirm events still fire after each removal.

    Evidence

    • PageSpeedGoogle PageSpeed (November 19, 2026): mobile 93, desktop 87; real visitors pass Core Web Vitals (field LCP 1.5 s, CLS 0.04). The homepage is 3.8 MB over 115 requests, 3.2 MB from 12 third-party hosts including Google Tag Manager, Hotjar, a HubSpot form connector, OptinMonster and Leadfeeder; render-blocking requests cost about 1 s; images could save 655 KB and unused JavaScript 628 KB.Source
  • 20

    Width already promotes an AI system eval ebook, "eval" is a stage of every agentic build, and the IPG case study details its quality controls (schema enforcement, LLM-based validation, tracing and hallucination screening). Many companies shipped agents in 2025-2026 without measuring them, and that is a smaller, faster purchase than a new build.

    High impactMedium effort First result in 6 weeks Lead AI Engineer

    First action

    Write the audit scope and price from the eval ebook's framework.

    How to measure it

    Eval audits sold

    Now
    –
    Target
    3 a quarter, half converting
    By
    2 quarters

    Steps

    1. 1

      Define scope

      One system, two weeks, fixed deliverables.

      Lead AI Engineer 2 days

    2. 2

      Price it

      Fixed fee.

      Founder & Chief Architect 1 hour

    3. 3

      Build the scorecard

      Accuracy, hallucination, latency, cost per task.

      Lead AI Engineer 1 week

    4. 4

      Page

      Offer page linked from agentic and RAG pages.

      Webflow Developer 2 days

    5. 5

      Offer to ebook leads

      Email eval ebook downloaders.

      Marketing Lead 1 day

    6. 6

      Run the first three

      Discounted pilots for case studies.

      Lead AI Engineer 6 weeks

    7. 7

      Convert

      Propose fixes as follow-on projects.

      Founder & Chief Architect ongoing

    First result
    6 weeks
    Cost
    Staff time; pilots discounted
    Owner
    Lead AI Engineer

    Smallest test

    Offer three pilot audits to existing contacts.

    Why it works

    Package a two-week "Agent eval audit" at a fixed price: Width reviews one production agent or RAG system, builds a test set from real traffic, measures accuracy, hallucination and cost per task, and delivers a scorecard with fixes. Offer it to eval ebook downloaders and in the newsletter's agents track, and link it from the agentic and RAG service pages. Audit findings become the scope for follow-on builds.

    A fixed-price audit is easy to approve, shows Width's engineering depth on the client's own system, and often turns into a fix-and-improve project.

    Risks and how to handle them

    • Client data access is slow

      Start from logs and a sample export under NDA.

    Evidence

    • Your websiteSitemap (266 URLs, checked November 19, 2026): 189 blog posts; service pages for chatbots, RAG, agentic AI, prompt engineers, NLP and RAG for healthcare; ebook pages for agentic RAG, AI system evals, content automation, AI for ecommerce and an Upwork playbook; no n8n service page and no post on project cost; Webflow template pages (/template-info/style-guide, /changelog, /licenses), /blog-old and two Heatrr.ai pages still listed.Source
    • Your website"Agentic AI for E-Commerce at Enterprise Scale" (June 22, 2026): an Amazon listing system for Interpublic Group with "roughly fourteen ReAct agent nodes" across 3,000 brands; it reports no measured result (time saved, accuracy or sales lift), only that it replaces "days of manual research".Source
    • Your websiteAgentic AI services page: "Your scoping call this week - POC in 31 days"; Hire-a-Dev, project-based and consulting models; proof missing from the homepage: a voice agent used by "10,000+ customers", a 14-node agent system for a holding company serving 3,000 brands, "96%+ field accuracy" for a medical-records startup and a 30% cut in sales team hours; visitors "leave the first call with a clear scope and the chatbot development cost"; no prices shown.Source

Big bets 3

Larger plays that can change the trajectory.

  • 21

    Width sells 31-day POCs and already offers Hire-a-Dev (an engineer in the client's team) and project builds, but the site presents them as separate choices. Rivals lead with ongoing models: Markovate's Forward Deployed Engineers and Addepto's managed delivery. Without a planned next step, a successful POC can end as a one-off invoice.

    High impactHigh effort First result in 1 quarter Founder & Chief Architect

    First action

    Add a "production options" section to the next POC proposal.

    How to measure it

    POC-to-retainer rate

    Now
    –
    Target
    40%
    By
    2 quarters

    Steps

    1. 1

      Define the plans

      Hire-a-Dev and a managed plan with fixed monthly scope.

      Founder & Chief Architect 1 week

    2. 2

      Update proposals

      Production options in every POC proposal.

      Founder & Chief Architect 1 day

    3. 3

      Build the handover pack

      Eval report, production plan, options.

      Lead AI Engineer 2 weeks

    4. 4

      Monitoring kit

      Reusable dashboards for managed clients.

      Lead AI Engineer 1 month

    5. 5

      Publish

      Show the model on the POC page.

      Marketing Lead 1 day

    6. 6

      Track

      Conversion and retention by client.

      Founder & Chief Architect monthly

    First result
    1 quarter
    Cost
    Staff time; monitoring tooling
    Owner
    Founder & Chief Architect

    Smallest test

    Offer the managed plan at the end of the next two POCs.

    Why it works

    Design the day-31 handover as a sale: every POC ends with an eval report, a production plan and two options, a Hire-a-Dev engineer or a managed plan (monitoring, evals and monthly improvements at a fixed fee). Write it into the POC proposal from day one, track POC-to-retainer conversion, and publish the model on the POC page.

    Recurring revenue smooths a project business, and the team that built the POC is the cheapest, lowest-risk choice to run it in production.

    Risks and how to handle them

    • Team capacity tied up in retainers

      Hire against signed retainers, not forecasts.

    Builds on

    Evidence

    • Your websiteAgentic AI services page: "Your scoping call this week - POC in 31 days"; Hire-a-Dev, project-based and consulting models; proof missing from the homepage: a voice agent used by "10,000+ customers", a 14-node agent system for a holding company serving 3,000 brands, "96%+ field accuracy" for a medical-records startup and a 30% cut in sales team hours; visitors "leave the first call with a clear scope and the chatbot development cost"; no prices shown.Source
    • Competitor researchRivals' own homepages: LeewayHertz, "AI development company enabling innovation and rapid development", with dedicated-team, team-extension and project models; Markovate, "300+ projects delivered", a scoped pilot of about 4-6 weeks and Forward Deployed Engineers; Addepto, "100+ AI projects" and managed delivery. None shows prices.Source
  • 22

    The homepage shows an AWS Partner Advanced Tier badge, AWS Solutions Architect and Azure AI certifications, a Pinecone Partner badge and n8n partner badges, but none of these relationships appears as a channel: there are no partner pages, no marketplace listings mentioned and no joint case studies.

    High impactHigh effort First result in 2 quarters Founder & Chief Architect

    First action

    Review which partner benefits the AWS Advanced Tier already unlocks.

    How to measure it

    Partner-sourced opportunities

    Now
    –
    Target
    4 a quarter
    By
    3 quarters

    Steps

    1. 1

      Audit benefits

      What each partner program offers today.

      Founder & Chief Architect 1 week

    2. 2

      Choose two

      Best customer overlap.

      Founder & Chief Architect 1 day

    3. 3

      Package offers

      POC and eval audit, partner-ready.

      Founder & Chief Architect 2 weeks

    4. 4

      Solution pages

      One per partner on width.ai.

      Content Writer 2 weeks

    5. 5

      Joint case study

      One per partner.

      Content Writer 1 month

    6. 6

      Meet partner teams

      Introductions and deal registration.

      Founder & Chief Architect ongoing

    7. 7

      Track

      Pipeline by partner.

      Marketing Lead monthly

    First result
    2 quarters
    Cost
    Staff time; possible marketplace fees
    Owner
    Founder & Chief Architect

    Smallest test

    Publish one AWS solution page and ask the partner manager for one introduction.

    Why it works

    Pick the two partners whose customers match Width best (AWS for enterprise agent builds, Pinecone for RAG). For each, publish a joint solution page, list a fixed-scope offer (the 31-day POC or the eval audit) where the partner program supports it, write one joint case study, and register opportunities with partner sales teams. Measure partner-sourced pipeline separately.

    Partner teams look for capable implementation firms to hand deals to, and a packaged offer with proof is what gets a small firm referred.

    Risks and how to handle them

    • Partner programs are slow for small firms

      Lead with one narrow, proven offer per partner.

    Builds on

    Evidence

    • Your websiteHomepage proof: testimonials from WorkOtter, Denison University, SalesHawk and JDM Garages ("Width.ai increased our advertiser revenue 36% in just that first month"); "over 50 services companies have worked with us"; badges for n8n Expert Partner, "Top Ranked n8n Agency", AWS Partner Advanced Tier, Azure AI Certified and Pinecone Partner; the "Everything ML" newsletter.Source
    • Your websiteSitemap (266 URLs, checked November 19, 2026): 189 blog posts; service pages for chatbots, RAG, agentic AI, prompt engineers, NLP and RAG for healthcare; ebook pages for agentic RAG, AI system evals, content automation, AI for ecommerce and an Upwork playbook; no n8n service page and no post on project cost; Webflow template pages (/template-info/style-guide, /changelog, /licenses), /blog-old and two Heatrr.ai pages still listed.Source
  • 23

    Width has repeated, specific accuracy results on catalogue tasks: 97% top-level and 93% six-level-deep on 5,585 categories, 92% on a multilingual wholesale catalogue, 97% five-level-deep, 89% SKU image classification and 92.44% product similarity. Few firms publish comparable numbers, and catalogue buyers have no neutral yardstick.

    Medium impactHigh effort First result in 1 quarter Lead AI Engineer

    First action

    Choose the public dataset and the three tasks to score.

    How to measure it

    Benchmark leads

    Now
    –
    Target
    20 qualified leads and 30 referring domains
    By
    3 quarters

    Steps

    1. 1

      Pick dataset

      Public product data with a taxonomy.

      Lead AI Engineer 1 week

    2. 2

      Define tasks

      Categorization, attributes, matching.

      Lead AI Engineer 1 week

    3. 3

      Run baselines

      Common approaches and Width's.

      Lead AI Engineer 1 month

    4. 4

      Publish code

      Scoring script and splits.

      Lead AI Engineer 1 week

    5. 5

      Results page

      Charts and method notes.

      Content Writer 2 weeks

    6. 6

      Launch

      Newsletter, LinkedIn and ecommerce communities.

      Marketing Lead 2 weeks

    7. 7

      Refresh

      Twice a year.

      Lead AI Engineer 6 months

    First result
    1 quarter
    Cost
    Engineering time and compute
    Owner
    Lead AI Engineer

    Smallest test

    Publish categorization results on one public dataset first.

    Why it works

    Build a public benchmark on an open product dataset (for example Google Product Taxonomy labels on public product data): categorization depth by depth, attribute extraction and matching, with Width's methods, Pumice and common baselines scored the same way. Publish the dataset split, the scoring script and a results page, and update it twice a year. Use it in sales, PR and the catalogue hub.

    A benchmark makes Width the reference point in catalogue AI, earns links and citations, and backs the 97% claim with something anyone can rerun.

    Risks and how to handle them

    • Width does not top every task

      Publish honestly and explain where custom work helps.

    Builds on

    Evidence

    • Your websiteCase studies page: 19 case studies; 13 date from 2021 to 2023. The newest are "Agentic AI for E-Commerce at Enterprise Scale" (June 22, 2026), a product-similarity study at 92.44% (July 16, 2026) and retail-shelf SKU recognition at 89% (August 30, 2026), plus three product-categorization studies at 92-97% accuracy (December 2024 to March 2025).Source
    • Pumice.aiPumice.ai, "Powered by Width.ai": "Ai Agents for SKU Onboarding & Product Data Enrichment", categorization at "over 97% accuracy", catalog PDFs to CSV and channel copy for Amazon, Walmart and eBay; no named customers; main CTA "Request a Demo".Source

Wildcards 1

Unexpected experiments, each with a cheap way to test it.

  • 24

    Width's free workflows (Google review replies, Square + Mailchimp, newsletter and graphic generators) live on GitHub, and each is mentioned inside a long post. Searches for a specific automation ("n8n google review reply workflow") would land better on a short page per workflow, and those pages would build width.ai's authority instead of GitHub's.

    Medium impactMedium effort First result in 2 months N8n Automation Engineer

    First action

    List every free workflow Width has published.

    How to measure it

    Workflow library traffic

    Now
    –
    Target
    3,000 a month
    By
    2 quarters

    Steps

    1. 1

      Inventory

      All free workflows on GitHub.

      n8n Automation Engineer 1 day

    2. 2

      Template

      Webflow collection for workflow pages.

      Webflow Developer 2 days

    3. 3

      Write pages

      Purpose, nodes, setup, download.

      Content Writer 2 weeks

    4. 4

      Record setups

      Two-minute video each.

      n8n Automation Engineer 1 week

    5. 5

      Link up

      From posts, the n8n page and GitHub READMEs.

      Webflow Developer 1 day

    6. 6

      Submit to gallery

      Best workflows to n8n templates.

      n8n Automation Engineer 1 week

    7. 7

      Add monthly

      One new workflow a month.

      n8n Automation Engineer monthly

    First result
    2 months
    Cost
    Staff time only
    Owner
    N8n Automation Engineer

    Smallest test

    Publish the Google reviews workflow as its own page.

    Why it works

    Create /n8n-workflows: one page per workflow with what it does, a diagram, the nodes and credentials needed, a one-click JSON download (still free), a setup video, and "Want it customised?" linking to the n8n service page. Add new workflows from client work (anonymised) each month and submit the best to n8n's template gallery with a link back.

    A library compounds: each page targets a specific, high-intent search, collects optional emails, and shows the depth behind the n8n partner badges.

    Risks and how to handle them

    • Workflows break when n8n updates

      Show a tested-on version and recheck quarterly.

    Builds on

    Evidence

    • Your website"Automate Google Reviews With AI Replies (Free n8n Workflow)" (November 19, 2026): the workflow downloads from GitHub with no email capture; CTAs offer free setup and "Talk to us about your reviews"; the "Keep Reading" section is empty and no case study is linked.Source
    • Your website"How to Automate Restaurant Email Marketing: From someone who has done it 17 times" (November 19, 2026): free Square + Mailchimp workflows on GitHub, a "Get my free setup" button and "You only pay if you find value"; no restaurant client is named.Source
    • Your websiten8n tag page: seven posts, including "5 n8n Marketing Automation Use Cases We Run for Real Customers" and "n8n vs LangGraph for Production AI Workflows", with no n8n service offer, only "Get In Touch" and a newsletter signup.Source
More example playbooks and reviews

More growth playbooks

Other companies we have researched, each read from what they are hiring for.

  • Genea growth strategyA cloud property technology platform: Genea Security (cloud access control, mobile keys, visitor management, Genea MCP), On-Demand HVAC…21 growth ideas
  • Care Access growth strategyA clinical research company that runs research sites, mobile clinics and the Future of Medicine community health screening program, and…22 growth ideas
  • G&H Automation growth strategyA systems consultancy that designs, cleans up and maintains CRM, automation and integration systems: HubSpot onboarding and monthly plans…24 growth ideas
  • Virtualitics growth strategyAI-native readiness software for defense: Virtualitics Iris and readiness applications for maintenance, materiel, personnel, supply chain…21 growth ideas
  • Rivington Solutions growth strategyA workflow-redesign consultancy that studies how one important business workflow really runs, redesigns it, and decides where AI or…22 growth ideas
  • Bild AI growth strategyAI takeoffs for Division 8: Bild AI reads plans and specs, extracts door schedules and hardware sets, sizes hardware, flags…22 growth ideas
  • The Animal Rescue Site growth strategyA free daily Click to Give site, funded by sponsor advertising, that pays for food and care for shelter pets. It sits alongside a Shopify…22 growth ideas
  • The Run Digital growth strategyA marketing agency for home service businesses in Quebec (the "Home Run Digital formula"): websites built to get calls, a lead capture…21 growth ideas

See every playbook and website review

Independent research. Programmable wrote this playbook from public sources, as they stood on November 19, 2026, and cites them throughout. Width.ai did not commission, review or endorse it, and Programmable is not affiliated with Width.ai. Company and product names are trademarks of their owners, used only to identify them. The ideas are suggestions and opinion, not professional, legal or financial advice, and not statements about how Width.ai runs its business.

Something wrong, or do you represent Width.ai? Email [email protected] and we will correct it or take it down.

30/60/90 roadmap

What to focus on in each month, and the ideas that carry it.

  1. Days 1 to 309 ideas

    Trust and hygiene: a real about page, current proof and a clear H1 on the homepage, homepage clean-up, a qualifying contact form, lead capture on free workflows, robots.txt and template pages, Heatrr retirement and accessibility.

  2. Days 31 to 6011 ideas

    Pipeline: an n8n service page, a published POC price band, measured case studies, a cost guide, the ecommerce catalogue hub, a Pumice sample run, a local-business automation package, a separate freelancer funnel, newsletter segments, script consolidation and an eval audit offer.

  3. Days 61 to 904 ideas

    Bets: POC-to-retainer, AWS and Pinecone co-selling, a public catalogue-AI benchmark and an on-site n8n workflow library.

The research behind the plan

2 research areas, each summarised in a few lines. Open any of them for the full detail.

Customers

What customers say

Customers value real AI expertise. The most cited problem: manual catalogue, listing and document work that does not scale.

  • Fast shipping: "No powerpoints - we just ship" and 31-day POCs
  • Not knowing what an AI project will cost or whether it will work
  • In their words: “POC”

Market

Competitive landscape

4 competitors compared: LeewayHertz, Markovate, Addepto and n8n freelancers and agencies.

  • Gap to exploit: Published prices for an AI proof of concept
  • LeewayHertz: Broad, generic positioning
  • Markovate: No public prices

From the team that wrote this playbook

All products
  • ProgrammableAgencyWorking AI systems in days, not quarters (opens programmable.agency in a new tab)
  • ProgrammableInfoAn agent that finds, researches and asks firstNew accounts get 100,000 free enrichment credits with code BETA100K (opens programmable.info in a new tab)
  • ProgrammableDigitalScreenshots in, an animated product demo out (opens programmable.digital in a new tab)
  • ProgrammableWatchKnow the day a supplier falls out of compliance (opens programmable.watch in a new tab)
  • ProgrammableTalkGet registrants for your events, and know who’s coming (opens programmable.talk in a new tab)
  • ProgrammableMediaAn AI note taker, interpreter and webinar host for your calls (opens programmable.media in a new tab)
  • ProgrammablePageText what you need, your customer gets a page that works (opens programmable.page in a new tab)
  • ProgrammableCommunityForums with labelled AI members, never empty (opens programmable.community in a new tab)
ProgrammableMarketing
ExamplesLeaderboardsProductsTermsPrivacyCookiesDPALegalContact

© 2026 AI-Powered Growth Intelligence